Fifty-six articles met our inclusion criteria .The great majority of the included articles were published in Chinese(n=53).Our results reveal various ethical challenges related to the sharing of biomedical data in China . We created an alluvial diagram to specify the result. Due to the limited amount of data processed by RAWGraphs at one time, the results of documents 1-28 and 29-56 are shown in two graphs respectively. We create four different steps. The first step is the number of literatures we collected. The second step is Year. In the case the size of each node is proportional to the number of rows that containing that specific value, it reveals that a growing body of literature on the biomedical data in the time period of 2011-2019.The largest flow of related published literature is in 2018.The third one is Research Method, and contains four nodes(case study, document analysis, questionnaire survey, interview). The fourth one is Ethical challenges, and contains five nodes, capacity building; interoperability: balancing different stakeholders’ interests; data subjects’ rights; scientific and social value; public trust and engagement. The largest flow is capacity building and interoperability: balancing different stakeholders’ interests ranked second. The flow among nodes in the four steps represent the number of lines sharing the same couple of values. In combination of two graphs, most documents in 2018 pay attention to capacity building. Case study are often presented in coordination with capacity building, while questionnaire surveys mainly focus on the data subjects’ rights and public trust and engagement.
Thematic analysis (table 1) identified a number of ethical challenges in the five overarching domains. These categories describe a landscape of challenges that interconnected and dynamic.
Capacity building
There is a wide spectrum of literature concern about the capacity building for sharing biomedical data. The challenge of capacity building was mainly discussed from the following three aspects: (1) isolated information island. Data was fragmented distributed in different institutions and data are often analyzed in institutional silos.
[2][9][12][25]For example, most hospitals in China use their own intranet respectively, which involves dozens of manufactures and each manufacturer has its own schemas(e.g.,data templates).[1][5]There is no data sharing platform that can support nationwide data sharing.[18][23][43] (2)Administrative sections’ poor overall planning abilities. The biomedical data sharing projects are short of long-term and integral plans, resulting in poor expansibility and interactivity systems. Regulatory systems were not designed nor updated to foster widespread collaboration. It is suggested more investment in platforms that can standardize and curate data into the uniform formats are required for sharing data effectively.[50] (3) Insufficient supporting policy for the training programs of biomedical and health informatics. Many biomedical scientists and physicians do not have any familiarity with the ideas of data science. There are not enough training programs for biomedical scientists to narrow the gap between needed and existing biomedical data science. skills.[12][20][41][49]
Interoperability: balancing different stakeholder’ interest
The financial, organizational, technical structures and data resources vary considerably across regions. While some developed regions benefit from sufficient financial support and solid infrastructure, other less developed regions do not have a sustainable basis for the data sharing activities. Even they collect biomedical data,they do not have the skills and expertise to analysis. Their sharing activities may be put at risk by participation (eg.by deepen the gap). As such, both the distribution and the amount of benefits are important. According to our review, data sharing activities in China are mainly driven by the following two forces. One is executive order driven. Under the requirements of govern agency, the data controllers are forced to share their data. The other is active data sharing. Data controllers actively took part in the sharing activates under the great potential of biomedical big data.
For the executive order driven data sharing activities, although government has issued a serious of policies, China is lack of specific laws and ethical guidelines to balance the competing considerations. For example, some sources states that there is no law that clearly defines the ownership of data and benefit sharing. The subject of responsibility is not clear either.[10][20][48]
The active data sharing activities also face challenges. Many articles report there are limited incentives for researchers and physicians to share data. They treat data as private property, and keep the data they obtained confidential and do not willing to share.[3][5][6][9] Other studies point out the sharing activity cannot bring benefits to the researchers in the short term. The benefit brought by data sharing is quite limited. And the sharing activity made their research more transparent and injure their interest. The data controllers also concern about the inherently unknowably variables of data sharing. Researchers may be reluctant to share data they spend years to collect because they worry about the rewards and recognition not be equitably shared within the corporators.
Social benefits and value
Maximize public health benefits and social value are the fundamental justification for data sharing. The typical argument is that collaboration and sharing allow for more effective analysis of the massive datasets and quicker translation of research findings into clinical practice and government decision. According to the various studies we collected, however, the utility rate of the biomedical data is still too low .For instance, report that the utility of the data is low because the data quality and validity have not been established. Raw data large and difficult to integrated. [58]
Patients’ data has not been widely shared between different hospitals, which resulted in repeated examination and bring burden to the patients. [11][20][28][43] Doctors are unable to analyze patient’s conditions comprehensively and accurately because of the lack of the complete diagnosis and treatment history of patients. And these data have not been integrated into clinical diagnosis and treatment. Biomedical data has not been integrated into public health services and infectious disease monitoring.[3][8][11][20]
The data subjects’ rights
In general, most included studies indicate that the data subject’s right is lack of attention and there is a huge risk of data abuse. Some studies states that the reason of it is hard to protect data subject’s right is because the government do not have specific regulation towards data sharing.[22][24]Some sources state that the data subjects’ right cannot be supported using the existing administrative and legal system. The data subjects do not have ownership of their data and it is easily gotten out of their control. The process of using this data is not transparent and the data subjects even do not have accessibility to their data. The laws tend not to be robust enough to adequately address ethical issues related to data sharing that matter to data subjects and researchers.[4][10][12][20]
Many sources state that the management system cannot technically guarantee the security of data. Poor data protection system also limits the sharing scope. According to “Risk Investigation Report on Sensitive Data Leakage of Medical Internet Services” conducted by Tencent company, there are serious logical loopholes in the third-party medical service platforms accessed by domestic 3-A-Grade hospitals. These vulnerabilities can lead to the disclosure of patient information, including name, mobile phone number, ID number, home address, registration record, inspection report, hospital record, medical report, payment record, etc.The report pointed out that The sensitive ports of medical Internet assets are more open, and the core business assets are directly exposed to the outside world, reducing the technical barriers of illegal hacking and unauthorized access. [57]
The current informed-consent procedures may have researched their limits with an increased data-sharing demands. Institutions do not have a governance system to obtain authorization for future use of these data in research. Some research emphasize that the data subjects lack the sense of privacy protection. Patients do not consider informed consent and privacy protection as important. According to an interview with doctors in 52 hospitals in five prefecture-level cities: Guangzhou, Dongguan, Shenzhen, Foshan, and Qingyuan, patients have poor health literacy and lack of judgment ability. [12] On the other hand, physicians do not treat the rights and welfare of individuals from whom the data were collected as important.
Public trust and engagement
According to CIMOS, the potential participants and communities’ participatory process do not only involve informed consent process, but also include the design, development, implementation, and monitoring of research and in the dissemination of the results. It appears that reliance on informed consent have failed to secure trustworthiness and public engagement. It is tightly connected to transparency, the accessibility of information about management and accountability. However, many studies point out that the data sharing process is not transparent and the data subjects cannot even obtain their clinical data. [10][25][28] Ineffective governance system is also an important reason for data subjects’ poor engagement. The governance structure does not clearly outline the responsibilities of involved data processors. (eg. whether the secondary use has met the intended purpose) and management agency acquiesces in the utilization of patients’ biomedical data by third parties.[4]
In a questionnaire survey of 397 domestic hospital patients, some indicated a preference to share biomedical data with their physicians and government agencies. They were convinced that the physicians and the relevant government departments might be related to “public”. When referred to the biopharmaceutical companies, their willingness to share had dropped significantly. The patients were not willing to share their data with other hospitals, which might be due to a significant increase in risk as data flows to the hospital. Their preferred informed consent model was one-time authorization submission and confirmed it periodically.[44]
IRB approval for biomedical data sharing is lagging. None articles we reviewed have mentioned the importance of review and approval procedures by an independent REC in the context of respect for individuals. The research ethics committees are considered as the gatekeepers and can limit the occurrence of unethical data sharing activities. Yet they have not been given to enough attention.
Table 1 Ethical challenges of biomedical data sharing in China and related CIMOS guidelines
Item
|
CIMOS
|
Ethical challenges (times mentioned in articles)
|
Capacity building
|
Research infrastructure building and strengthening research capacity
|
Data fragmented distribution and low level of data standardization ,resulting in technical barriers (30)
|
Heterogeneous sources and multiple standards(8)
|
Insufficient information sharing mechanism and architecture (5)
|
Same system repeat construction, waste of resources, lack of long-term consideration(1)
|
The hospitals use the internal network to share data and this network could not connect with the external network, which makes sharing difficult. (4)
|
The raw data’s volume is massive and hard to integrate(2)
|
Poor data quality and incomplete content (9)
|
Shared content excessively single(1)
|
Developing technologies appropriate to health care and health-related research
|
The informatization level is low and still in the developing stage(5)
|
Educating research and health-care personnel
|
Insufficient supporting policy for the training programs of biomedical and health informatics (1)
|
Health IT talent shortage(16)
|
Doctors’ basic knowledge of medical informatics is weak, and their enthusiasm for learning are not high. (3)
|
Doctors are under high pressure and their incentives to participate are not high (1)
|
Interoperability: balancing different stakeholder’ interests
|
Data sharing requires careful balancing of competing considerations.
|
Unevenly distributed medical data resources and information network infrastructure between regions (19)
|
The coverage of shared subject is small, lacking data sharing projects covering the whole country (5)
|
Medical data resources are unevenly distributed in different hospitals in the same area (1)
|
Government has insufficient investment to the base medical units(7)
|
The level of Informatization in primary medical institutions is low, and it is difficult for patients to obtain quality medical resources. (1)
|
There is no law that clearly defines the ownership of data and the sharing of interests. (3)
|
Management agencies have poor management ability(8)
|
Numerous management entities, supervised by multiple departments (2)
|
Data sharing regulation system is not sound (7)
|
Researchers are unaware of the value of sharing (1)
|
No evaluation system (1)
|
Not enough cooperation between hospitals and enterprises (1)
|
Arranging for joint publication consistent with recognized authorship requirements and data sharing
|
Data controllers lack the sense of communication and cooperation, only pay attention to their hospitals/institutions’ interests, and keep the biomedical data confidential. (11)
|
Preparing a benefit-sharing agreement to distribute eventual economic gains from the research
Researchers, sponsors and research ethics committees must share data for future research where possible.
|
Not enough communication between data controllers (2)
|
Medical agencies are divided and concern their own interests(13)
|
Hospital administration office worried about information leakage, and have a very cautious attitude toward data sharing(2)
|
Sharing information does not bring benefits to hospitals in the short term (4)
|
Medical practices are more open and transparent, touching the interests of hospitals (4)
|
Unclear responsibility(1)
|
Scientific and Social value
|
The ethical justification for undertaking health-related research involving humans is its scientific and social value: the prospect of generating the knowleg4e and the means necessary to protect and promote people’s health
|
Patients have to repeat exanimate because biomedical data are not shared between hospitals, resulting in waste of time and money (1)
|
Data utilization is low and has not used for medical decision making, prevention and monitoring (4)
|
Low data utilization, not used for scientific research (1)
|
Poor data quality misleading patients (1)
|
Wearable devices data is idle and wasted.
(2)
|
Primary medical institutions cannot meet the needs of patients (3)
|
The data subjects’ rights
|
When data are stored,
institutions must have a governance system to obtain authorization for future use of these data in research
|
Management system cannot guarantee the security of data (1)
|
The existing legal system cannot effectively protect the rights of data owners (7)
|
The government acquiesces in the development and utilization of patient data by third parties (1)
|
Patients do not have ownership of their biomedical data, the data is out of the control (3)
|
Excessive collection of data (1)
|
Government departments have not made clear provisions on residents' health information sharing (2)
|
Custodians of the data must arrange to protect the confidentiality of the information linked to the data,by sharing only anonymized or coded data with researchers, and limiting access to the material of third parities
|
Information theft and data tampering (1)
|
Doctors have little awareness of protecting patients' privacy (1)
|
Data subjects' privacy is at risk (4)
|
Technology cannot guarantee the security of data and faces privacy risks (3)
|
Patients have little awareness of privacy (4)
|
Seek and obtain consent, but only after providing relevant information and ascertaining that the potential participant has adequate understanding of the material facts
|
Poor interoperability, asymmetric doctor-patient information (3)
|
Citizens have low health literacy and lack of ability to safeguard their rights (3)
|
The controllers share data without obtaining informed consent from the data subjects (4)
|
Public trust and engagement
|
researchers,
sponsors, health authorities and relevant institutions should engage in a meaningful. That involves them in an early and sustained manner in the design,
development,
implementation, design of the informed consent process and monitoring of research
|
It is still in the early stage of development and cannot guarantee the trust and benefit of data subjects (4)
|
Patients are reluctant to share their data (1)
|
The data is not transparent, and the data subjects cannot obtain their clinical data (1)
|
Strengthening research ethics review and oversight capacity in host communities
|
It is difficult for data subjects to protect their rights and find responsible parties (2)
|
One-way information sharing, lack of interaction with data subjects (2)
|